!
BARRELS 2025
San Diego, California
Organizers
Solange Brown (Johns Hopkins)
Josh Brumberg (CUNY)
Randy Bruno (Oxford)
Dan Feldman (Berkeley)
Mitra Hartmann (Northwestern)
Kate Hong (CMU)
Dieter Jaeger (Emory)
Krishna Jayant (Purdue)
David Kleinfeld (UCSD)
Keerthi Krishnan (UTK)
Soohyun Lee (NIH/NIMH)
Farzaneh Najafi (Georgia Tech)
Daniel O'Connor (Johns Hopkins)
Simon Peron (NYU)
Carl Petersen (EPFL)
Scott Pluta (Purdue)
Robert Sachdev (Humboldt-Berlin)
Gordon Shepherd (Northwestern)
Jochen Staiger (Goettingen)
Edward Zagha (UC Riverside)
Click here for link to vibrissa.org
BARRELS 2025
San Diego, California
BARRELS 2025 page 1
THURSDAY MORNING: 8:00 AM - 1:00 PM
8:00 AM – 9:00 AM BREAKFAST AND COFFEE
9:00 AM – 9:05 AM Welcome: Randy Bruno, Oxford University
9:05 AM – 9:35 AM Invited Talk: Distinct projection-specific L5B dendritic reorganization as a
function of value-based learning. Krishna Jayant, Purdue University.
9:35 AM – 10:05 AM Invited Talk: Cortical circuit dynamics during learning and memory. Lucy
Palmer, University of Melbourne.
10:05 AM – 10:15 AM STRETCH BREAK
10:15 AM – 10:30 AM Short Talk: Structured and target-specific development of cortico-cortical
connectivity in the mouse visual cortex. Matthew W. Jacobs, John M. Ratliff,
Alec L.R. Soronow, Jordan A. Nichols, Hylen T. James, Jorin A.G. Eddy, Adam
M. Murray, and Euiseok J. Kim, University of California, Santa Cruz.
10:30 AM – 11:00 AM Invited Talk: Neural circuits for flexible auditory learning. Kishore
Kuchibhotla, Johns Hopkins University.
11:00 AM - 11:15 AM Short Talk: Intrinsic interval timing, not temporal prediction, underlies
ramping dynamics in visual and parietal cortex, during passive behavior.
Yicong Huang, Ali Shamsnia, Mengze Chen, Shuang Wu, Tim Stamm, Sophie
Medico, and Farzaneh Najafi, Georgia Institute of Technology.
11:15 AM - 11:30 AM Short Talk: Impact of eye movements and orofacial movements on mouse
visual cortex. Atika Syeda, Miguel Nunez, Lin Zhong, Marius Pachitariu, and
Carsen Stringer, Janelia Research Campus, HHMI.
11:30 AM - 11:45 AM COFFEE BREAK
11:45 AM - 12:15 PM Invited Talk: Neocortical mechanisms of visual perceptual inference.
Hyeyoung Shin, Seoul National University.
12:15 PM - 12:30 PM Short Talk: Emergence of tonotopically organized spontaneous activity in
the brain after genetically induced hearing loss. Patrick D. Parker, Riley T.
Bottom, Ulrich Müller, and Dwight E. Bergles, Johns Hopkins University.
12:30 PM - 12:45 PM Short Talk: Microglial-orchestrated remodeling of perineuronal nets in
response to sensory deprivation. Katherine A Rivera Gómez, Victoria Bamwo,
Ezekiel Willerson, and Joshua C Brumberg, The Graduate Center, and Queens
College of the City University of New York.
12:45 PM - 1:00 PM Short Talk: High-speed voltage imaging of action potentials in molecular
layer interneurons reveals sensory-driven synchrony that augments
whisker movement. Brown ST, Holla MR, Land MA, Yang S, McDonald AJ, St-
Pierre F, and Raman IM, Northwestern University.
BARRELS XXXIX, 2025
San Diego, California
BARRELS 2025 page 2
THURSDAY AFTERNOON 2:30 PM - 5:00 PM
1:00 PM – 2:30 PM LUNCH BREAK
2:30 PM - 2:45 PM Short Talk: Local and Global Neural Dynamics Underlying Stimulus and
Choice Across Sensorimotor Cortex. Andrew S. Blaeser, Mitchell Clough,
Allison M. Ahrens, and Jerry L. Chen, Boston University.
2:45 PM – 3:15 PM Invited Talk: Inhibitory and excitatory specificity from a connectomics
census in mouse visual cortex. Nuno da Costa, Allen Institute.
3:15 PM – 3:45 PM Invited Talk: Connecting single-cell transcriptomes to the projectomes in
mouse visual cortex. Staci Sorensen, Allen Institute.
3:45 PM - 4:00 PM COFFEE BREAK
4:00 PM - 4:30 PM Invited Talk: A transcriptomic and epigenomic cell type atlas of the
developing mouse visual cortex. Hongkui Zeng, Allen Institute.
4:30 PM - 4:45 PM Short Talk: Imaging voltage in parvalbumin interneurons in slices of mouse
somatosensory cortex. Meyer Jackson, University of Wisconsin.
4:45 PM - 5:00 PM Short Talk: Natural scene statistics in touch: hands and whiskers. Neeli
Tummala, Adrienne L. Fairhall, and Mitra J. Hartmann, Northwestern University
and the University of Washington.
5:00 PM – 8:00 PM DINNER AND POSTERS
BARRELS XXXIX, 2025
San Diego, California
BARRELS 2025 page 3
FRIDAY MORNING: 8:00 AM – 2:00 PM
8:00 AM – 9:00 AM COFFEE
9:00 AM - 9:30 AM Invited Talk: Multi-modal representations in thalamus and cortex. Randy
Bruno, Oxford University.
9:30 AM - 10:00 AM Invited Talk: The brainstem trigeminal nucleus represents tactile events in
multiple reference frames. Jesse Goldberg, Cornell University.
10:00 AM - 10:15 AM STRETCH BREAK
10:15 AM - 10:30 AM Short Talk: Cell-type-specific genetic variation associated with individual
differences in goal-directed learning. Alanna E. Carey, Halley L. Dante, Kevin
M. Delgado, David G. Lee, Rhea Singh, Noah Tan, Daisy E. Powers, Fani Memi,
Kenny Roberts, Omer A. Bayraktar, Jerry L. Chen. Boston University and the
Wellcome Sanger Institute.
10:30 AM – 10:45 AM Short Talk: X-ray based wire-by-wire vibrissa to barrelette mapping. Matias
Mugnaini, Ben Gerhardt, and Michael Brecht, Humboldt-Universität zu Berlin,
and the Max Planck School of Cognition, Leipzig.
10:45 AM - 11:00 AM Short Talk: Population codes supporting adaptive sensorimotor learning in
the posterior dorsal striatum. C. Ivan Linares, Sofia E. Juliani, Jessie Yi, and
David J. Margolis, Rutgers University.
11:00 AM – 11:15 AM Short Talk: Perceptual learning enhances the temporal integration of self-
motion and touch in somatosensory cortex. Hyein Park, Hayagreev Keri and
Scott Pluta. Purdue University. Missing abstract
11:15 AM – 11:30 AM COFFEE BREAK
11:30 AM - 11:45 AM Short Talk: Reactivated thalamocortical plasticity alters neural activity in
sensory-motor cortex during post-critical period. Hyesoo Jie, Emily Petrus,
Nikorn Pothayee, and Alan P Koretsky, No Affiliation.
11:45 AM – 12:00 PM Short Talk: Flexible temperature-light touch interactions across subcortical
somatosensory circuits. Anda M. Chirila and David D. Ginty. Brown University.
12:00 PM - 12:15 PM Short Talk: Different functional neuronal systems modulate the brain’s own
tactile input. Cornelius Schwarz, Ritu Roy Chowdhury, Kalpana Gupta, Yuyao
Sun, Fransiska Gekeler, and Shubhodeep Chakrabarti. University Tuebingen,
12:15 PM - 12:30 PM Short Talk: Lateral spread in sensory cortex as a novel gain mechanism in
selective detection. Angelina Lam and Edward Zagha, University of California
Riverside.
12:30 PM – 2:00 PM LUNCH
END OF MEETING
BARRELS XXXIX: Abstracts for Invited Talks
BARRELS 2025 page 4
Abstracts for invited talks are listed in alphabetical order
by last name of presenting author
Invited talks
1. Randy Bruno, Oxford University. Multi-modal representations in thalamus and cortex.
Virtually every cortical area is associated with its own primary and high-order thalamic nuclei. While primary nuclei are well
understood to relay sensory information from the periphery to the cortex, they account for little of thalamus. Thalamus is
mainly comprised of high-order nuclei, whose functions have been elusive. Theoretical roles for high-order sensory nuclei
have spanned complex sensory processing, motor efference copy aggregation, egocentric representation, and spatial- and
feature-based attention. Recent rodent studies from my lab and others, which I will discuss, have not found strong evidence
for these theories. Most recently, we leveraged comparisons of high-order somatosensory (PO) and visual thalamus
(pulvinar, LP) and found these two nuclei are highly synchronized despite being interconnected with separate sensory
modalities. Both nuclei appear to encode the occurrence of salient or behaviorally relevant events, regardless of any
associated stimulus modality. High-order thalamus may exist to enhance cortical plasticity to facilitate learning of behavioral
tasks.
2. Nuno da Costa, Allen Institute. Inhibitory and excitatory specificity from a connectomics census in mouse visual cortex.
The mammalian cortex comprises a rich diversity of neuronal cell types, each defined by distinct anatomical, molecular, and
functional characteristics. Synaptic connectivity determines how these cell types integrate into cortical circuits, yet mapping
connectivity rules at the resolution of individual cell types remains a major challenge. Here, we employed millimeter-scale
volumetric electron microscopy to examine the synaptic connectivity of inhibitory and excitatory neurons across a densely
segmented population spanning all layers of the mouse visual cortex. This approach enabled the construction of a
comprehensive wiring diagram of excitation and inhibition. Drawing inspiration from classical neuroanatomy, we classified
inhibitory neurons based on their dendritic compartment targeting, and developed a novel classification of excitatory neurons
using dendritic reconstructions paired with whole-cell maps of synaptic input. Our analysis revealed widespread specificity
in inhibitory connectivity onto excitatory neurons, with many interneurons selectively targeting spatially intermingled
excitatory subpopulations. Inhibitory targeting was organized into "motif groups"—diverse sets of interneurons that
collectively innervate both perisomatic and dendritic compartments of shared excitatory targets. We further investigated
how excitatory neurons interact with these inhibitory motif groups and with other excitatory cells. Similar to inhibitory
connectivity, excitatory connectivity exhibited extensive specificity across both supragranular and infragranular layers. In
addition, we identified a recurrent circuit motif: intratelencephalic (IT) neurons, regardless of subtype or laminar position,
target excitatory partners both directly and indirectly via inhibitory neurons. We propose that this motif modulates the timing
and gain of postsynaptic excitatory responses, constraining firing windows and preventing runaway excitation.
3. Jason Gao, Brendan Ito Jesse Goldberg, Cornell University. The brainstem trigeminal nucleus represents tactile events
in multiple reference frames.
Accurate goal-directed movements require information about effector position and ongoing motion to be integrated into
sensory feedback. Here, with high-speed videography, we examined three-dimensional tongue kinematics as mice drank
from a water spout and re-aimed their licks using subtle tactile events on the tongue induced by random spout displacement
to the left, center, or right. Because a tactile event on the left side of the tongue can specify a centered spout on right licks,
or a left spout on straight licks, task performance requires a coordinate transformation from a tongue- to head-based
reference frame. In past work, we showed that mice integrate information about both precise touch events and tongue
position to re-aim ensuing licks, and that the superior colliculus (SC) contains a sensorimotor map for touch-guided lick
aiming (Ito, Gao et al., 2025). Importantly, SC neurons encode tactile events in both reference frames, suggesting that the
SC implements a coordinate transformation or inherits these signals from its inputs. Yet another possibility is that the SC
inherits these diverse representations from inputsI from lower levels of the sensory hierarchy. To test this idea, we recorded
neurons in the brainstem trigeminal nucleus (SpV), the first relay center for tongue tactile inputs into the brain with
projections going to SC. Surprisingly, SpV neurons could represent tacti
BARRELS XXXIX: Abstracts for Invited Talks
BARRELS 2025 page 5
4. Shulan Xiao1, Lautaro F. Soler1, John Morris1, Andres E. Uzcategui1, Saumitra Yadav1, Krishna Jayant1, 2 1. Purdue
University, Weldon School of Biomedical Engineering, West Lafayette, IN, USA 2. Purdue University, Purdue Institute for
Integrative Neuroscience, IN, USA Distinct projection-specific L5B dendritic reorganization as a function of value-based
learning
Value-based learning in the barrel cortex likely requires flexible coordination of sensory and motor signals through
projection-specific dendritic computations in layer 5B pyramidal tract neurons (L5BPNs). These thick-tufted neurons
integrate feedforward input from basal dendrites with top-down feedback on their apical tufts, yet how their integration rules
adapt with learning remains unclear. Using dual somato-dendritic patch-clamp recordings ex vivo, and two-photon calcium
imaging of retrogradely labeled barrel cortex L5BPNs in vivo, we reveal projection-specific differences in dendritic coupling
and morphology that govern integration modes. Pons-projecting neurons exhibit deeper apical bifurcations, stronger back-
propagation, and resonance-like bursting with efficient soma–apical coupling, whereas PoM- and SC-projecting neurons
show pronounced apical compartmentalization and coincidence detection requiring synchronous input. During an active
two-whisker discrimination task, two-photon imaging revealed learning-dependent remodeling of apical activity patterns,
marked by enhanced coactivation and dynamic restructuring in PoM- and SC-projecting subspaces, but to a lesser degree
in PONs-projecting. Together, these results demonstrate that value-based learning drives distinct, projection-dependent
dendritic reorganizations that modulate coupling efficiency and computation across L5BPN subtypes. Such adaptive
reconfiguration of dendritic integration provides a cellular mechanism linking learning value to hierarchical sensorimotor
coordination across cortical and subcortical circuits. Funding: This work was supported by the following grants to K.J. the
National Institutes of Health, New Innovator Award NIH DP2MH136494. This material is also based upon work supported
by the Air Force Office of Scientific Research under award numbers FA9550-23-1-0701. Any opinions, findings, conclusions,
or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the
United States Air Force. This work is also supported by EMBRIO Institute, contract #2120200, a National Science
Foundation (NSF) Biology Integration Institute.
5. Kishore Kuchibhotla, Johns Hopkins University. Neural circuits for flexible auditory learning
I will present recent work demonstrating that the auditory cortex (AC) is necessary for discrimination learning but
dispensable at expert levels. Rather than being driven by tonotopic map expansion or tuning shifts, distinct higher-order
computations—reward prediction for learning and action suppression for performance—govern behavioral output and are
carried by spatially clustered neural ensembles within the AC. These clusters are uncoupled from stimulus-tuning,
revealing a higher-order organizational principle of the AC that extends beyond sensory representations. I will then outline
future directions in my lab aimed at understanding the neural mechanisms supporting lifelong, multi-task learning, bridging
insights from learning theory, the functional architecture of neural circuits, and computational principles of compositionality
and generalization.
6. Lucy Palmer, University of Melbourne. Cortical circuit dynamics during learning and memory.
Central to learning and memory is the ability of the cortex to dynamically encode and store learned information through
changes in neural activity and connectivity. While dendrites are the primary sites of synaptic input, their role in learning
and memory processes remains poorly understood. To address this, we used two-photon calcium imaging from both layer
2/3 (L2/3) pyramidal neuron dendrites, as well as perirhinal axonal projections, in frontal cortex during a closed-loop
visuomotor choice wheel task. Following learning, calcium signals in both dendrites and perirhinal axonal inputs were
selectively enhanced during correct task performance. Disrupting the closed-loop behaviour altered this signalling,
significantly changing the frequency of evoked calcium events. These findings reveal that the perirhinal-frontal cortical
microcircuit not only supports the transfer of learned information but is also influenced by memory disruption. Together,
these findings provide new insight into how long-term memories are represented and maintained in cortical networks,
highlighting a key role for dendritic and perirhinal signalling in associative learning.
BARRELS XXXIX: Abstracts for Invited Talks
BARRELS 2025 page 6
7. Hyeyoung Shin, Seoul National University. Neural Code of Perceptual Inference.
Perception is a process of inference, whereby incoming sensory evidence is interpreted based on prior expectations about
the sensory world. Thus, the neural code of perception should be evaluated based on how well it implements optimal
perceptual inference. However, the neural code of perception has conventionally been evaluated by its capacity to
accurately represent sensory information. I argue that due to this misalignment in the computational goal of perception,
assessments of the neural code have been biased towards categorization over generalization, and efficiency over
robustness. Taken together, the neural code should be evaluated based on how well it facilitates the goal of perception, i.e.,
perceptual inference. This work was supported by the Samsung Science and Technology Foundation (SSTF-BA2302-07),
the National Research Foundation of Korea(NRF) grant funded by the Korea government(MSIT) (RS-2024-00358070, RS-
2024-00413689, RS-2023-00301976), and the Seoul National University New Faculty Startup Fund.
8. Staci A. Sorensen*1, Nathan W. Gouwens*1, Yun Wang*1, Matt Mallory1, Agata Budzillo1, Rachel Dalley1, Brian Lee1,
Olga Gliko1, Hsien-chi Kuo1, Xiuli Kuang5, Rusty Mann1, Ed Lein1, Jim Berg1, Brian Kalmbach1, Shenqin Yao1, Hui
Gong2,3, Qingming Luo4, Lydia Ng1, Uygar Sumbul1, Tim Jarsky1, Zizhen Yao1, Bosiljka Tasic1, and Hongkui Zeng1
*These authors contributed equally to this work 1Allen Institute for Brain Science 2 Britton Chance Center for Biomedical
Photonics, Wuhan National Laboratory for Optoelectronics, MoE Key Laboratory for Biomedical Photonics, Huazhong
University of Science and Technology, Wuhan, China 3HUST-Suzhou Institute for Brainsmatics, JITRI Institute for
Brainsmatics, Suzhou, China 4State Key Laboratory of Digital Medical Engineering, School of Biomedical Engineering,
Hainan University, Haikou, China 5School of Optometry and Ophthalmology, Wenzhou Medical University, Wenzhou, China
Connecting single-cell transcriptomes to projectomes in mouse visual cortex.
The mammalian brain consists of diverse neuron types with different functions. Recent single-cell RNA sequencing
approaches led to a whole brain taxonomy of transcriptomically-defined cell types. Patch-seq experiments augment these
cell type descriptions by linking transcriptomic profiles with local morphological and electrophysiological properties.
However, linking transcriptomic identities to long-range axonal projections remains a major unresolved challenge. To
address this, we collected a coordinated data set in mouse visual cortex consisting of excitatory Patch-seq neurons, with
morphological, electrophysiological, and transcriptomic data collected from the same cell, and excitatory, whole neuron
morphologies (WNMs). From the Patch-seq data, we defined 17 integrated morpho-electric-transcriptomic (MET)-types and
built a multi-step classifier to integrate cell type assignments with WNM and interrogate cross-modality relationships. Layer
5 neurons displayed the greatest diversity across all modalities, with nine of the excitatory MET-types found within this
population. We find that transcriptomic variations within and across MET-types correspond with morphological and
electrophysiological phenotypes. In addition, this variation, along with the anatomical location of the cell, can be used to
predict projection targets of individual neurons. Funding: NIH U19MH114830, 1RF1MH128778-01, and Allen Institute
founders, P. G. Allen and J. Allen
9. Hongkui Zeng, Allen Institute. A transcriptomic and epigenomic cell type atlas of the developing mouse visual cortex.
The mammalian cortex is composed of a highly diverse set of cell types and develops through a series of temporally
regulated events. We report a comprehensive and high-resolution transcriptomic and epigenomic cell type atlas of the
developing mouse visual cortex. The atlas was built from a single-cell RNA-sequencing dataset and a single-nucleus
Multiome dataset, densely sampled from E11.5 to P56. We constructed a transcriptomic developmental trajectory map of
all excitatory, inhibitory, and non-neuronal cell types in the visual cortex. The trajectory map shows that neurogenesis,
gliogenesis and early postmitotic maturation in the embryonic stage gives rise to all cell classes and nearly all subclasses
in a staggered parallel manner. Increasingly refined cell types emerge throughout the postnatal differentiation process,
including during eye opening and the onset of critical period, suggesting continuous cell type diversification at different
stages of cortical development. Throughout development, we find cooperative dynamic changes in gene expression and
chromatin accessibility in specific cell types, identifying cell-type specific and temporally resolved gene regulatory networks
linking transcription factors and target genes through accessible chromatin motifs. Our study provides a real-time dynamic
molecular map associated with specific cell types and temporal events that can reveal the molecular logic underlying the
multifaceted cortical cell type and circuit development.
BARRELS XXXIX: Abstracts for Short Talks
BARRELS 2025 page 7
Abstracts for short talks are listed in alphabetical order by
last name of presenting author
Short talks
1. Andrew S. Blaeser, Mitchell Clough, Allison M. Ahrens, Jerry L. Chen. Boston University, Dept. of Biology Local and
Global Neural Dynamics Underlying Stimulus and Choice Across Sensorimotor Cortex.
Animals often encounter the same stimulus but respond differently based on goals. How the brain shifts between goal- and
non-goal-directed behavior moment-to-moment involves complex sensorimotor interactions. Here, mice were trained in a
whisker-based task requiring flexible responses to the same tactile stimuli. Optogenetic inactivation of somatosensory or
motor cortex altered behavior regardless of goal context. Using simultaneous two-photon imaging of four sensorimotor
regions, we analyzed local versus global activity patterns encoding task information and inter-areal communication.
Stimulus, choice, and their associations were represented in S1, S2, and M1. In S1 and S2, task and communication
subspaces were distinct, while in M1, choice-related subspaces aligned with communication channels during goal-directed
phases. Top-down flow from motor to sensory areas was essential for correct choices. These findings highlight that flexible
sensorimotor communication supports adaptive, stimulus-driven behavior. Funding: NIH U01MH10907, NIH R01NS140230
2. Brown ST, Holla MR, Land MA, Yang S, McDonald AJ, St-Pierre F, and Raman IM. Northwestern University, Evanston,
IL. High-speed voltage imaging of action potentials in molecular layer interneurons reveals sensory-driven synchrony that
augments whisker movement.
Testing whether the synchrony of action potential firing is a cerebellar coding mechanism requires simultaneous recording,
with high temporal fidelity, from populations of identified neurons. Here, we used targeted one-photon voltage imaging at 2-
4 kHz to record action potentials from groups of ∼10-300 molecular layer interneurons (MLIs) expressing a positively tuned,
genetically encoded voltage indicator, FORCE1f or pAce. In awake resting mice, crus I MLIs fired brief (∼1-ms) spikes at
20-60 spikes/s. Sensory stimuli of air puffs to the whiskers evoked short-latency (<10 ms) increases in spiking probability.
In most trials, >50% of MLIs fired synchronously with 4-ms temporal precision. The magnitude of puff-evoked whisks
correlated tightly with the trial-by-trial percentage of synchrony. Brief optogenetic stimulation of MLIs was sufficient to induce
and augment whisker protraction, whereas overriding MLI inhibition by stimulating target Purkinje cells reduced protractions,
providing direct evidence that sensory-evoked spike synchrony can generate movement. Supported by NIH grant R35-
NS116854.
3. Alanna E. Carey1,2*, Halley L. Dante 2,3, Kevin M. Delgado1, David G. Lee2,3, Rhea Singh1, Noah Tan1, Daisy E.
Powers1, Fani Memi4, Kenny Roberts4, Omer A. Bayraktar4, Jerry L. Chen1,2,3. 1 Department of Biology, Boston
University, Boston, MA. 2 Neurophotonics Center, Boston University, Boston, MA. 3 Department of Biomedical Engineering,
Boston University, Boston, MA. 4 Wellcome Sanger Institute, Wellcome Genome Campus, Hinxton, Cambridge, UK. Cell-
type-specific genetic variation associated with individual differences in goal-directed learning.
Individuals exhibit variability in their ability to learn complex tasks. We investigated whether genetic variation associated
with learning could manifest as gene expression differences in specific cell types. We assayed mice on a 6-week sensory-
guided learning task using a large-scale automated homecage training system. Over 350 mice of different genetic
backgrounds were tested including DO mice and their eight founder inbred lines. We observed heritable differences in task
learning. The strain (NZO) were consistently able to achieve expert task performance while all individuals from two other
strains (PWK, NOD) failed to do so. To identify genes associated with learning, QTL, eQTL, and DEG analyses were
performed on DO mice. 531 learning associated genes were identified across the brain. To determine if these candidate
genes converged onto specific cell types, we performed spatial transcriptomics on naïve and trained NZO, PWK, and NOD
animals. Using latent factor analysis, we found differentially expressed candidate genes between learner and non-learner
strains including in oligodendrocyte and striatal interneuron subtypes. Additional cell types, including glutamatergic cortical
neurons, were identified that showed training-induced gene expression changes irrespective of task learning. These findings
show individual differences in task learning are associated with cell-type-specific gene expression patterns that are both
inherited and experience-dependent. Funding: NIH F99
BARRELS XXXIX: Abstracts for Short Talks
BARRELS 2025 page 8
4. Anda M. Chirila1 and David D. Ginty2,3 1 Department of Neuroscience, Brown University, Providence, RI 02912 2
Department of Neurobiology, 3 Howard Hughes Medical Institute, Harvard Medical School, Boston, MA 02115. Flexible
temperature-light touch interactions across subcortical somatosensory circuits.
Functionally distinct thermal and mechanosensory afferents converge in the spinal cord dorsal horn, yet how dorsal horn
circuits combine them to shape perception remains poorly understood. We used large-scale in vivo electrophysiological
recordings in anesthetized mice and a range of well-controlled tactile and thermal stimuli applied to the skin to characterize
sensory coding in laminae I–V dorsal horn interneurons and across distinct projection neuron populations. Innocuous and
noxious cooling and heating were represented in a distributed manner that extends far beyond the thermoreceptor-recipient
laminae I and II. Analysis of 653 single units revealed thermal thresholds tiling the spectrum of thermal stimuli from 0 °C to
50 °C. Hierarchical clustering of thermal and touch-evoked responses revealed that 76.3% of dorsal horn interneurons are
polymodal, encoding temperature and innocuous mechanical force with distinct adaptation kinetics and indicating extensive
convergence of mechanoreceptor and thermoreceptor input. Despite this extensive convergence at the population level,
optotagged projection neurons showed segregated encoding. These findings position the dorsal horn as a dynamic
integrator that non-linearly transforms temperature and touch inputs into parallel, functionally distinct ascending outputs.
5. Meyer Jackson, University of Wisconsin Imaging voltage in parvalbumin interneurons in slices of mouse somatosensory
cortex.
Hybrid voltage sensors (hVoS) are genetically encoded and can thus be targeted to neurons based on their genetic identity.
This provides a powerful general approach to the study of neural circuit mechanisms. By crossing an hVoS Cre reporter
with various Cre drivers we have targeted >10 different cell types and have imaged their electrical activity in various brain
regions. Here a parvalbumin (PV) Cre driver was used to target probe. In slices from mouse somatosensory cortex we could
image voltage responses to electrical stimulation in >50 PV interneurons simultaneously. Response amplitude and half-
width varied between neurons residing in L4 and L2/3. Furthermore, stimulating L2/3 elicited responses with longer latencies
than stimulating L4. Using the responses of PV interneurons to track propagation in excitatory axons revealed conduction
velocities ranging from 74 to 473 um/msec; interlaminar conduction was more rapid than intralaminar conduction. Focused
stimulation of single excitatory neurons elicited unitary synaptic responses in PV interneurons; stimulating single stellate
cells elicited larger responses than stimulating single pyramidal cells. The RNA binding protein FXR1 has been implicated
in mental illness, as have PV interneurons. Targeted deletion of FXR1 from PV interneurons resulted in larger response
amplitudes and shorter latencies. Thus, PV interneurons have diverse response characteristics that can be modulated by
FXR1. Funding: NIH grant R35 NS127219.
6. Matthew W. Jacobs1,3, John M. Ratliff1,3, Alec L.R. Soronow1, Jordan A. Nichols1, Hylen T. James1, Jorin A.G. Eddy1,
Adam M. Murray1, Euiseok J. Kim1,2,* 1Department of Molecular, Cell, and Developmental Biology, University of
California, Santa Cruz, CA 95064 2Institute for the Biology of Stem Cells, University of California, Santa Cruz, CA 95064,
USA 3Co-first author, *Corresponding author Structured and Target-Specific Development of Cortico-Cortical Connectivity
in the Mouse Visual Cortex.
The mammalian cortex shows highly stereotyped long-range connectivity, yet the developmental principles that specify
precise cortico-cortical projection patterns remain unclear. Two models propose that target specificity arises either from
early exuberant outgrowth followed by pruning or from directed initial axonal targeting. To address this, we mapped postnatal
development of V1 cortico-cortical projection neurons (CCPNs) to eleven higher visual areas (HVAs) in mice using rapid
retrograde, anterograde, and single-cell tracing methods. We found that V1→HVA connectivity develops via staggered axon
extension and pruning aligned with target position along the medial-lateral axis. Reciprocal feedback from HVAs to V1 was
also refined, yielding aligned bidirectional connectivity. Both multiplexed retrograde tracing and MAPseq-based single-cell
profiling showed that individual V1 neurons initialize and retain specific projection motifs with limited variation, arguing
against global exuberance followed by selective pruning. Instead, our findings support a directed guidance model in which
distinct V1 CCPN subtypes establish selective projection patterns early, followed by local target-dependent refinement. This
structured yet heterogeneous strategy provides an anatomical framework for emergence of precise long-range cortical
networks.
BARRELS XXXIX: Abstracts for Short Talks
BARRELS 2025 page 9
7. C. Ivan Linares, Sofia E. Juliani, Jessie Yi, David J. Margolis; Department of Cell Biology and Neuroscience, Rutgers.
Population codes supporting adaptive sensorimotor learning in the posterior dorsal striatum.
The tail of the striatum (TS) is implicated in auditory–motor integration and decision-making, yet how TS spiny projection
neurons (SPNs) reorganize their activity as animals learn sound–action mappings remains incompletely understood. We
performed longitudinal two-photon calcium imaging in head-fixed mice while they learned a joystick manipulandum-based
behavior that requires a forelimb push or pull in response to auditory cue tones. Population activity aligned to movement
onset showed only weak correlation, but strong modulation on correct versus incorrect trials. The proportion of modulated
neurons overall and in both D1 and D2 populations, and the tuning strength of modulated neurons for sound, action, reward,
or mixed, increased with performance. With task complexity, mice increased reliance sound cues rather than prior reward
or action. With reversal of the sound-action mapping, a subset of neurons represented the new correct sound-action
mapping and other neurons transiently retained signatures of the previous rule. Collectively, our results show that D1 and
D2 SPNs track learned sound–action associations with performance-dependent strengthening and sharpening of
modulation, enhanced reward encoding, and a code that is functionally stable at the subcategory level yet flexible at the
single-neuron and pathway levels. These findings shed new light on adaptive population codes in the TS during
sensorimotor learning.
8. Matias Mugnaini 1; Ben Gerhardt 1, 2; Michael Brecht 1, 3; 1 Bernstein Center for Computational Neuroscience,
Humboldt-Universität zu Berlin, Germany; 2 Max Planck School of Cognition; Leipzig, Germany; 3 Neurocure Cluster for
Excellence; Berlin, Germany. X-ray based wire-by-wire vibrissa to barrelette mapping.
Despite exquisite understanding of multi-vibrissa projections onto barrelette-, barreloid-15and barrel-maps, connectivity
principles of single-vibrissa afferents onto single brainstem barrelettes remain elusive. Here, we combined a novel follicle-
nerve-brainstem preparation and X-ray imaging at multiple synchrotron facilities to connect C2-vibrissa and brain wire-by-
wire. Vibrissa afferents evenly sample deflection angles by relatively uniform innervation around the vibrissa and leave the
follicle in an ordered angularly linear arrangement. Afferents become less20ordered in nerve and ganglion, but retain some
angular topography. In the brainstem, afferents dispatch two main-branches, setting up two repeating, ordered, linear
angular maps along the anterior-posterior barrelette axis. Backward deflection angles – activated in vibrissal touch – are
greatly overrepresented. The C2 principalis-barrelette is a linearized representation of vibrissa deflection angles selectively
relaying backward deflections to thalamus. M.M. receives funding by Human Frontiers Science Program. B.G., M.M.and
M.B. are funded by BCCN Berlin. M.B. is supported by Neurocure Cluster of Excellence.
9. Yicong Huang, Ali Shamsnia, Mengze Chen, Shuang Wu, Timothy Stamm, Sophie Medico, Farzaneh Najafi Intrinsic
interval timing, not temporal prediction, underlies ramping dynamics in visual and parietal cortex, during passive behavior.
Neural activity following regular sensory events can reflect either elapsed time since the previous event (temporal signaling)
or temporal predictions and prediction errors about the next event (temporal predictive processing). These mechanisms are
often confounded, yet dissociating them is essential for understanding neural circuit computations. We addressed this by
performing two-photon calcium imaging from distinct cell types (excitatory, VIP and SST) in layer 2/3 of visual (VIS) and
posterior parietal cortex (PPC), while awake mice passively viewed audio-visual stimuli under temporal contexts with
different inter-stimulus interval (ISI) distributions. Computational modeling revealed distinct functional clusters of neurons,
including stimulus-activated (ramp-down) and stimulus-inhibited (ramp-up) categories, with distinct kinetics and area/cell-
type biases. Importantly, all functional clusters were invariant to temporal predictability, shifted immediately when temporal
statistics changed, and were identical between naive and experienced mice. Population decoding revealed that clusters
with heterogeneous kinetics differed in how well they represented interval information, such that together they tiled elapsed
time and produced a distributed, learning-independent population code for time. These results provide strong evidence
against temporal predictive processing in Vis/PPC under passive conditions and instead demonstrate intrinsic coding of
interval timing, redefining the mechanistic origin of ramping and omission-related activity in sensory cortex. We discuss how
these dynamics align with stimulus-reset attractor frameworks, and propose that temporal predictive processing is more
likely implemented in other circuits or recruited in Vis/PPC during task-engaged behavior.
BARRELS XXXIX: Abstracts for Short Talks
BARRELS 2025 page 10
10. Patrick D. Parker1, Riley T. Bottom1, Ulrich Müller1, Dwight E. Bergles1,2,3 (1The Solomon H. Snyder Department of
Neuroscience, Johns Hopkins University, Baltimore, MD, USA 2Department of Otolaryngology-Head and Neck Surgery,
Johns Hopkins University, Baltimore, MD, USA 3Kavli Neuroscience Discovery Institute, Johns Hopkins University,
Baltimore, MD, USA) Emergence of tonotopically organized spontaneous activity in the brain after genetically induced
hearing loss.
Hearing loss can induce secondary disorders, such as tinnitus, but our understanding of the centrally generated neural
patterns that underlie phantom percepts after sensory loss is limited. We developed an inducible model of rapid and
complete hearing loss by genetically disabling hair cell mechanoelectrical transduction (MET) in 4–8-week-old mice (Tmie
cKO mice), and defined the spatiotemporal characteristics of spontaneous neural activity in the brain using in vivo wide-
field and multiphoton Ca2+ imaging. MET inactivation resulted in the complete loss of sound-evoked responses within seven
days, and was accompanied by the emergence of large-scale, coordinated sound-independent (SI) neural activity patterns
in both the inferior colliculus and auditory cortex. These SI events aligned with the same isofrequency domains evoked by
pure tones in control (hearing) mice in both regions, reflecting the simultaneous firing of groups of adjacent neurons that
used to process the same sound frequency. SI activity in the colliculus of Tmie cKO mice was independent of the cochlea
but required synaptic input from ascending afferents. Using MET inactivation as a model of inducible sensory loss, we find
that spontaneous neural activity is coordinated along sensory domains, creating tonotopic-like patterns in lower-order
brainstem regions that propagate through the auditory system independent of sound.
11. Katherine A Rivera Gómez1,3, Victoria Bamwo3, Ezekiel Willerson2,3, Joshua C Brumberg1,2,3 1Graduate Center of
the City University of New York, 1Biology and 2Psychology, New York, NY, 3Queens College of the City University of New
York, Psychology, Queens, NY. Microglial-Orchestrated Remodeling of Perineuronal Nets in Response to Sensory
Deprivation.
Sensory deprivation (SD) during the critical period of neural development triggers microglial-mediated degradation of
perineuronal nets (PNs), a specialized extracellular matrix structure that preferentially surrounds Parvalbumin-positive (PV+)
interneurons, which are essential for the maturation and function of cortical inhibitory circuits. Sensory deprivation (SD)
elicits morphological, phagocytic, and transcriptomic shifts in microglia, key modulators of synaptic refinement and network
homeostasis. We targeted macrophage/microglia response markers in extracellular matrix remodeling using RNAScope
and Immunohistochemistry. Following SD, expression of tissue plasminogen activator (tPA), a serine protease implicated
in PV+ interneuron migration and perineuronal net (PN) proteolysis, was upregulated. Legumain (Lgmn), a lysosomal
cysteine protease, seen in matrix remodeling via metalloproteinases, such as MMP9, which is secreted by microglia that
target the main component of PNs was upregulated following SD with greater impact in female animals. The expression of
Aquaporin 4 (Aqp4), an astrocytic membrane protein with implications in astrocyte-microglia communication, is upregulated
in female SD animals, suggesting a role in astrocytic involvement during PN remodeling. These findings highlight microglial
involvement in PN degradation and suggest microglia-astrocyte collaboration may promote plasticity in the barrel cortex
after neonatal sensory deprivation.
12. Cornelius Schwarz, Ritu Roy Chowdhury, Kalpana Gupta, Yuyao Sun, Fransiska Gekeler, Shubhodeep Chakrabarti.
Systems Neuroscience, Center for Integrative Neuroscience, Hertie Institute for Clinical Brain Research University
Tuebingen, Germany. Different functional neuronal systems modulate the brain’s own tactile input.
The brain is thought to house distinct predictive systems that span from sensorimotor to cognitive domains. In this report, I
focus on the functional differentiation of two putative predictive mechanisms at the neuronal level: state estimation (SE) and
sensory gating (SG)—both of which serve to attenuate sensory input during movement. By examining sensorimotor
neuronal responses across the depth of the primary somatosensory cortex in mice trained on a whisker-based reach task,
SE emerged as a learned suppression of tactile signal transmission. This suppression was driven by a predictive signal that
coincided precisely with the expected sensory consequence of the movement. In contrast, SG persisted over a much longer
period, beginning after the initiation of a motor command and extending well beyond it. Both SE and SG appear to be
internal predictive processes, as their effects remained intact even when sensory feedback (reafference) was blocked. We
propose that SG may be linked to higher-level cognitive monitoring of movement goals, whereas SE likely reflects the
classical concept of the reafference principle.Funding: This research was supported by grants from the Deutsche
Forschungsgemeinschaft (SCHW577/21-1, SCHW577/21-2, SCHW577/22-1).
BARRELS XXXIX: Abstracts for Short Talks
BARRELS 2025 page 11
13. Atika Syeda, Miguel Nunez, Lin Zhong, Marius Pachitariu, Carsen Stringer. Impact of eye movements and orofacial
movements on mouse visual cortex.
Previous studies have found that the mouse primary visual cortex (V1) is correlated to various orofacial movements.
However, recent work in primates suggests that monkey V1 is primarily modulated by eye movements, not orofacial
movements. In mice, it remains unclear how much eye movements contribute to the modulation of neural responses in the
presence or absence of visual input compared to other orofacial movements. To determine the contribution of eye
movements to mouse V1 activity, we recorded the activity of thousands of V1 neurons using two-photon calcium imaging
while presenting a visual stimulus or in darkness and monitoring eye movements and orofacial behaviors with a camera.
We utilized Facemap, our AI-based framework, to precisely track orofacial movements and predict neural activity from
mouse behavioral videos. With this model, we found that eye movements influence a small fraction of activity in the mouse
visual cortex compared to other orofacial behaviors. In the presence of visual input, eye position predicts ~10% variance
explained after accounting for retinal input whereas orofacial behaviors account for 20-35% of variance explained. We
developed a convolutional neural network to predict eye movements from orofacial movements. We found whisking and
sniffing behaviors have more information about eye movements than vice versa. This suggests that orofacial movement
signals play a larger role in visual cortical processing in mice compared to primates.
14. Neeli Tummala(1); Adrienne L Fairhall(2); Mitra J. Hartmann (1, 3) (1) Mechanical Engineering, Northwestern
University, Evanston, IL, USA; (2) Neurobiology and Biophysics, University of Washington, Seattle, WA, USA; (3)
Biomedical Engineering, Northwestern University, Evanston, IL, USA. Natural scene statistics in touch: hands and whiskers.
Sensory systems have evolved to exploit the statistical structure of their inputs. Natural scene statistics are well-studied in
vision and audition, but tactile statistics are less explored. We first quantified the “tactile prior,” i.e., the statistical structure
of surfaces independent of sampling, and are now examining how the prior is encoded by the rodent vibrissal system and
the human hand. We analyzed a large dataset of natural and human-made 3D object meshes, computing surface curvature
across multiple spatial scales. Natural objects showed heavy-tailed distributions, long-range correlations, and scale
invariance; these are features also seen in visual and auditory scenes. In contrast, human-made objects had more
pronounced features at distinct spatial scales and reduced scale invariance. These findings suggest that tactile systems
share coding principles with other senses. To investigate how the tactile prior might shape neural encoding, we are now
using WHISKiT (Zweifel et al., 2021) to simulate mechanical signals at the base of rodent whiskers, and Touchsim (Saal et
al., 2017) to simulate cutaneous mechanoreceptors responses in the human hand. Our methods and findings establish a
statistical foundation for diverse lines of inquiry in sensory neuroscience, haptics, and robotics. Thanks to NO Zweifel for
object dataset; OY Lee for extra object scans; ME Black for WHISKiT simulations. Support: NIH R01NS-116277 (MJH);
Simons Collab Global Brain (ALF).
15. Angelina Lam and Edward Zagha, University of California Riverside. Lateral spread in sensory cortex as a novel gain
mechanism in selective detection.
Neocortex is a context-dependent connectivity machine. And yet, the mechanisms underlying dynamic functional
connectivity within the neocortex are poorly understood. Here, we investigate the modulation of lateral spread within sensory
cortex as a potential gain mechanism in the context of a selective detection task in mice. In this task, mice learn to selectively
respond to a single paddle deflection in the whisker field of one side of the face (target) and ignore identical deflections of
the opposite side (distractor). Target and distractor stimuli evoke similar responses within their primary somatosensory
cortex (S1) center response fields. And yet, target stimuli effectively propagate beyond S1 to multiple cortical and subcortical
regions while distractor stimuli do not. Using widefield Ca 2+ imaging, we determined the extent of target vs. distractor
lateral spread within sensory cortex. We found that target stimuli evoke significantly larger volumes of activation compared
to distractor stimuli, which emerges with learning. Using a simplified neural circuit model, we demonstrate that modulations
of lateral excitation/inhibition can account for the differences in target vs distractor evoked spatial activations. From these
findings, we propose that modulation of lateral spread in sensory cortex could be a potent mechanism of stimulus gain
underlying dynamic functional connectivity. Funding: NIH R01NS107599
BARRELS XXXIX: Abstracts for Short Talks
BARRELS 2025 page 12
16. Hyesoo Jie 1, Emily Petrus 2, Nikorn Pothayee 3, Alan P Koretsky No Affiliation. Reactivated thalamocortical plasticity
alters neural activity in sensory-motor cortex during post-critical period.
Neuroplasticity in sensory brain areas supports adaptation after nerve injury and fundamentally impacts sensation and
movement. However, limited neuroplasticity in somatosensory areas due to the early critical period makes determining the
role of thalamocortical (TC) inputs in sensorimotor signal processing challenging. Here, we demonstrated that reactivation
of TC neuroplasticity was associated with an increase in the number of neurons in layer IV (L4) of the whisker primary
somatosensory cortex (wS1) with a stable excitation-inhibition ratio. Highly synchronized neural activity in L4 propagated
throughout the wS1 column and to the downstream areas, including whisker secondary somatosensory, primary motor
cortices, and contralateral wS1. These results provide crucial evidence that TC inputs can alter the neural activity of sensory-
motor pathways even after the critical period. Altogether, these enormous changes in sensorimotor circuit activity are
important for adaptation following an injury such as limb loss, stroke, or other forms of neural injury.
BARRELS XXXIX: Poster Abstracts
BARRELS 2025 page 13
Abstracts for short talks are listed in alphabetical order by
last name of presenting author
Posters
1. Gamze Güney1,2,3, Mikkel Vestergaard4, Mario Carta5, James FA Poulet1,2. 1. Department of Neuroscience, Max
Delbrück Center for Molecular Medicine (MDC), Berlin-Buch, Germany 2. Neuroscience Research Center, Charité-
Universitätsmedizin Berlin, Germany 3. Department of Biology, Humboldt-Universität zu Berlin, Germany 4. Department of
Neuroscience, University of Copenhagen, Denmark 5. CNRS, Interdisciplinary Institute for Neuroscience, University of
Bordeaux, France
Thermal encoding by GABAergic inhibitory interneurons in the posterior insular cortex.
The processing of sensory information by the neocortex is at the heart of conscious perception and is known to involve a
dynamic interaction between synaptically connected GABAergic inhibitory interneurons and excitatory pyramidal neurons.
One hypothesis is that sensory feature encoding in pyramidal neurons is shaped by different interneuron types performing
specific functional roles. Here we address this in the mouse thermal cortex. The thermal system is a fundamental pathway
required for accurate object identification, pain perception and body temperature regulation, and is one that is highly
developed in the mouse. Recently, we identified a region of the posterior insular cortex (pIC) that contains somatotopic
maps of thermal and tactile information and is required for non-painful thermal perception (Vestergaard et al., 2023; Bokiniec
et al., 2023). Two-photon imaging has shown that pIC pyramidal neurons have a fine-scale and dynamic encoding of skin
temperature, but, to date, there is no information on the encoding of temperature by cortical interneurons. Here I present
our recent two-photon imaging of interneuron subtypes during thermal stimulation of the forepaw in awake mice. pIC
interneurons respond to non-painful cooling and warming with strong and reliable responses. Future work aims to
understand the role of interneurons in thermal perception with the use of a two alternative forced choice thermal perception
task.ERC-2015-CoG-682422, FOR 2143, SFB 1315
2. John Judge, Meyer Jackson; Biophysics Program, Dept. of Neuroscience; U. of Wisconsin-Madison.
Biased inter-columnar communication and short-term plasticity in mouse barrel cortex
Barrel cortex (BC) input exhibits a phase-, direction-, and frequency-dependent structure arising from whisking kinematics
and thalamic nuclei processing. It is unclear how BC extracts relevant spatiotemporal features from this input. To investigate
communication within and between cortical barrels, we targeted a hybrid voltage sensor (hVOS) to Scnn1a excitatory
neurons in BC layer 4 (L4) of male and female mice (mean age 7.8 ± 0.4 weeks), and imaged population responses to
electrical stimulation. Coronal and sagittal slices presented well-delineated barrels in L4, aligned approximately to whisking
axes. Voltage imaging tracked activity along an L4→L2/3→L4 relay during inter-barrel communication. AMPA receptor
blockade demonstrated that this relay depends on excitatory synaptic transmission and revealed intra- and inter-barrel
feedforward inhibition. We found differences in latency and velocity between protraction- and retraction-related directions
and also identified direction-dependent synaptic depression shaping inter-barrel communication. While dense hVOS
labeling reveals population activity very clearly, we also developed an in silico model for subpopulation activity aimed to
extend our insight into how synaptic processes and subcellular structure contribute to behavior at the circuit and system
levels.Funding: NIH grant R35 NS127219, NIH training grant T32 GM130550
BARRELS XXXIX: Poster Abstracts
BARRELS 2025 page 14
3. Dongho Kang & Hyeyoung Shin. School of Biological Sciences, Seoul National University
Higher-order connectivity, jointly determined by inter-soma distance, net synaptic inputs and net synaptic outputs, influences
local recurrent dynamics in mouse V1 layer 2/3.
Understanding the logic of recurrent connectivity in neocortex and linking it to neural activity has been a long-standing goal
of neuroscience. The functional connectomics data spanning multiple areas of mouse visual cortex provides an
unprecedented opportunity to address these questions. Using MICrONS dataset, we sought to determine the rules
underlying higher-order connectivity in primary visual cortex (V1) layer 2/3. Connection probability between a pair of neurons
depends on the inter-soma distance. While this ‘distance rule’ is widely accepted for first-order connectivity, it is not sufficient
to recapitulate higher-order connectivity. We found that scaling the distance-based connection probability by the total
number of input and output synapses for each neuron succeeded in recapitulating second-order connectivity with
remarkable accuracy. Next, we sought to identify neural activity metrics that are indicative of connectivity measures. Prior
literature suggested that population coupling, defined as correlation between a neuron’s activity and the population net
activity, is indicative of net synaptic inputs. We verified this relationship among proofread V1 layer 2/3 neurons in the
MICrONS data, and in the in-silico simulation of this network. In sum, higher-order connectivity is determined by net synaptic
inputs and outputs in addition to inter-soma distance, and higher-order connectivity influences the relationship between
population coupling and net synaptic inputs.
4. *M. R. Keaton, Max Planck Inst. for Neurobio. of Behavior, Bonn, Germany; M. Oberlaender, Integrative Neurophysiol.,
Vrije Univ. Amsterdam, Amsterdam, Netherlands. Predicting Structure-Function Relationships in Cortex via Artificial Neural
Networks
The coupling of information streams is a hallmark feature of cortical function, and it is believed that network architecture is
key to this. Neuroanatomical studies have shown that the specificity of projections of particular cell types both to and from
the cortex facilitate the formation of characterizable networks. However, determining the impact these structures have on
how incoming information streams are coupled and processed is not well understood and remains challenging to observe.
Here, we propose a computational approach to studying this by informing artificial neural network models with increasing
detail from neuroanatomical reconstructions of the cortex. We demonstrate that by training such cortically-inspired networks
on a battery of machine learning tasks, we obtain concrete predictions on how network architecture and wiring specificity
therein could facilitate function. We explore such structure-function relationships with respect to biologically-relevant tasks
like generalization and show how these networks compare to other possible architectures. Our approach provides promising
results and empirically testable predictions which we hope will shed new light on how interareal connectivity patterns
facilitate the coupling of information streams.
5. Jehyun Kim, Hyeyoung Shin School of Biological Sciences, Seoul National University, Seoul, South Korea. Neural
variability structure in primary visual cortex is optimal for robust representation of visual similarity
How different neuronal populations construct a robust representation of the sensory world despite neural variability is a
mystery. We found that neural variability in mouse primary visual cortex observe a simple rule: For a given sensory stimulus,
the mean and the variance of spike counts follow a linear relationship across neurons. To understand how this neural
variability structure affects the sensory representation, we artificially varied the slope of the log-mean and log-variance
relationship. We found that the intrinsic structure of neural variability allows representations of distinct sensory information
to be continuous while minimizing overlap, enabling the neural code to be robust while still being efficient. Further,
representational similarity was maximally consistent between different sets of neurons at slope 1, both within and across
mice. Thus, the intrinsic neural variability structure optimizes the robustness of the neural code, and may enable different
brains to build a common representation of the sensory world.Funding: This work was supported by the Samsung Science
and Technology Foundation (SSTF-BA2302-07), the National Research Foundation of Korea (NRF) grant funded by the
Korean government (MSIT) (RS-2024-00358070, RS-2024-00413689, RS-2023-00301976), and the Seoul National
University New Faculty Startup Fund.
BARRELS XXXIX: Poster Abstracts
BARRELS 2025 page 15
6. Alice Y. Nam1,2, Jiwook Shin1,2, Morgan Tenney1,2, Baihe Zhang1, and Y. Kate Hong1,2; 1 Department of Biological
Sciences, Carnegie Mellon University, Pittsburgh, PA 15213, USA 2 Neuroscience Institute, Carnegie Mellon University,
Pittsburgh, PA 15213, USA Somatosensory cortex shapes perceptual decisions via the superior colliculus
Perceptual decisions require integrating sensory evidence with internal criteria, but how cortical and subcortical circuits
coordinate this process is unclear. We investigated the roles of primary somatosensory cortex (S1) and superior colliculus
(SC) in mice performing a tactile detection task. Optogenetic suppression of contralateral S1 increased detection threshold,
whereas ipsilateral suppression decreased it. These effects reflected shifts in decision bias rather than decreased sensitivity.
Direct SC inactivation produced similar bias shifts, and SC lesions abolished S1-induced effects, confirming that S1
influences detection behavior primarily via downstream SC. Simultaneous recordings revealed that both S1 and SC encoded
whisker stimuli, but SC activity more reliably predicted choice and remained robust during S1 inactivation. Our results
identify SC as the key decision hub, with S1 biasing its output to shape detection behavior.
7. Maria Royo (1), Arco Bast (3), Rieke Fruengel (1), Christiaan P. J. de Kock (2), Marcel Oberlaender (1, 2) 1) In Silico
Brain Sciences Group, Max Planck Institute for Neurobiology of Behavior, Bonn, Germany 2) Dept. of Integrative
Neurophysiology, Center for Neurogenomics and Cognitive Research, Vrije Universiteit Amsterdam, Amsterdam, The
Netherlands 3) Janelia Research Campus, Howard Hughes Medical Institute, Ashburn, VA, USA. Broad receptive fields in
cortex facilitate efficient and robust population coding of sensory information
In conventional views of the cortical circuitry, sensory processing begins in layer 4 (L4), where inputs from primary thalamus
evoke responses that are selective to specific stimulus features. However, it has become increasingly clear that, in parallel,
the same thalamic input also drives layer 5 pyramidal tract (L5PT) neurons. In contrast to the narrow receptive fields (RFs)
in L4, L5PTs respond unselectively to a variety stimulus, having broad RFs. This raises the question: What information
about the sensory stimulus could L5PTs broadcast to their downstream targets despite their unspecific responses? To
address it, we combined realistic multi-scale models of the barrel cortex with electrophysiological measurements in-vivo.
We found that the broad RFs of L5PTs enable a population code that robustly broadcasts every sensory input to any
downstream subcortical area. Remarkably, even a small population of L5PTs—whether located within the same or across
different barrel columns—can encode the identity of the sensory stimulus. The reliability and efficiency of this code depend
on both the breadth and variability of the RFs. In essence, the broader and more diverse the RFs among L5PTs, the fewer
neurons are needed to convey the same sensory information. These results suggest that sensory inputs are encoded in
parallel by two complementary population codes: one based on the selective responses in L4, and another on the
unselective, variable responses of L5PTs
.
8. Lucas Williamson, Sukrith Vedapuram, Christina Dai, Cici Liang, Rowan Gargiullo, Gordon Berman, Chris Rodgers.
Coordination of whole body movement in mice
In daily life, we use both hands to complete many tasks like driving a car or buttoning a shirt. Deficits in bimanual coordination
present early in Parkinson’s Disease (PD) and resist treatment, causing disruption to everyday life. The precise patterns of
network activity that enable bimanual coordination—and how these patterns degrade in PD—remain unclear. To address
this unknown, I designed a behavior assay in which mice freely climb. As they climb, I collect kinematic and single unit
recordings from motor cortex using 3D tracking and wireless electrophysiology. The MotionMapper unsupervised machine
learning pipeline reveals that mice engage in many behaviors during exploration of the climbing arena, and classic tuning
curve analyses show individual neurons encode movement parameters across behavioral state. To study neural activity at
the population level, I use latent variable modeling to compute patterns of neural activity that encode individual forelimb
movement, which I hypothesize to occupy distinct subspaces of population activity in MOp and overlapping subspaces in
MOs. Next, I plan to compare these patterns of activity between healthy mice and MitoPark mice, a genetic model of PD.
After a period of symptom progression across 12-16 weeks of age, climbing behavior decreases in MitoPark mice before
vanishing entirely. I hypothesize that as deficits to bimanual coordination increase, patterns of MOs activity underlying
individual limb movement become more distinct.
